Purpose: We used gene microarray analysis to compare the global expression profile of genes involved in adaptation to training in skeletal muscle from chronically strength-trained (ST), endurance-trained (ET), and untrained control subjects (Con).
Methods: Resting skeletal muscle samples were obtained from the vastus lateralis of 20 subjects (Con n = 7, ET n = 7, ST n = 6; trained [TR] groups >8 yr specific training). Total RNA was extracted from tissue for two color microarray analysis and quantative (Q)-PCR. Trained subjects were characterized by performance measures of peak oxygen uptake (V˙O2peak) on a cycle ergometer and maximal concentric and eccentric leg strength on an isokinetic dynamometer.
Results: Two hundred and sixty-three genes were differentially expressed in trained subjects (ET + ST) compared with Con (P < 0.05), whereas 21 genes were different between ST and ET (P < 0.05). These results were validated by reverse transcriptase polymerase chain reaction for six differentially regulated genes (EIFSJ, LDHB, LMO4, MDH1, SLC16A7, and UTRN. Manual cluster analyses revealed significant regulation of genes involved in muscle structure and development in TR subjects compared with Con (P ≤ 0.05) and expression correlated with measures of performance (P < 0.05). ET had increased whereas ST had decreased expression of gene clusters related to mitochondrial/oxidative capacity (P ≤ 0.05). These mitochondrial gene clusters correlated with V˙O2peak (P < 0.05). V˙O2peak also correlated with expression of gene clusters that regulate fat and carbohydrate oxidation (P < 0.05).
Conclusion: We demonstrate that chronic training subtly coregulates numerous genes from important functional groups that may be part of the long-term adaptive process to adapt to repeated training stimuli.
1Department of Physiology, Monash University, Clayton, Victoria, AUSTRALIA; 2Centre for Aging, Rehabilitation, Exercise and Sport, Victoria University, Melbourne AUSTRALIA; 3Exercise Metabolism Group, School of Medical Sciences, RMIT University, Bundoora, Victoria, AUSTRALIA; 4Liggins Institute, University of Auckland, Auckland, NEW ZEALAND; and 5Victorian Bioinformatics Consortium, Monash University, Clayton, Victoria, AUSTRALIA
Address for correspondence: Nigel K. Stepto, Ph.D., Exercise Physiology Centre for Aging, Rehabilitation, Exercise and Sport, School of Sport and Exercise Science, Victoria University, PO Box 14428, Melbourne, Victoria, 8001, Australia; E-mail: firstname.lastname@example.org.
Submitted for publication October 2007.
Accepted for publication August 2008.
Skeletal muscle exhibits remarkable plasticity, demonstrated by its ability to adapt to various stimuli such as environmental stress, altered nutrient availability, and contractile activity (2,4,5,14,15,38). This adaptive capacity contributes to the marked differences in muscle phenotype (e.g., contractile velocity, mitochondrial content, and muscle cross-sectional area) that contribute to physical performance capacity.
Endurance and strength training represent divergent stimuli, inducing distinct changes that serve to minimize cellular stress in subsequent similar exercise sessions (9). Endurance training elicits metabolic and morphological responses (15,16), including mitochondrial biogenesis (1,21), a fast-to-slow twitch fiber phenotype transformation (38), a decreased reliance on carbohydrate-based fuels during submaximal exercise (20), and an improved whole-body aerobic capacity (V˙O2peak;). In contrast, strength training has minimal effect on mitochondrial biogenesis or patterns of substrate metabolism (35) but increases protein synthesis and hypertrophy that promote gains in maximal force output (23,27) when compared with endurance training (18). Notably, the exercise-induced adaptation to both endurance and strength training is an important factor for enhancing metabolic health and for maintaining functional capacity.
An acute bout of exercise stimulates numerous intramyocellular signals whose up-regulation lasts from seconds to hours after cessation of each contractile bout (5,14). These signals ultimately result in altered gene expression profiles that are specific to the training-induced stress (13,14,17,31,38). Subsequently, gene expression returns to preexercise levels until further contraction-induced perturbation occurs (25,37). Chronic adaptations to training likely represent the summation of many single (acute) exercise bouts with these perturbations leading to cumulative, global alterations in gene and protein expression and, ultimately, the distinct phenotypic changes seen after prolonged periods of strength or endurance training. It is possible that the expression of key genes or clusters of genes that contribute to the altered phenotype after long-term training programs are chronically altered in athletes compared with sedentary individuals. Such a model has been proposed by Flück (12), but the validity of this model has yet to be tested experimentally using global gene expression analysis or using subjects involved in strength training.
Recently, gene microarrays have been used to examine acute (22,31,36,37) and chronic (19,24,28,33) changes in gene expression profiles of healthy, aging, and/or diseased skeletal muscle in an attempt to identify the underlying genetic alterations responsible for certain disease states. Such studies have been successful in identifying differential gene expression patterns in these populations. The aim of the present study was to use gene microarray technology to compare the gene expression patterns in skeletal muscle of chronically endurance-trained (ET) and strength-trained (ST) subjects, using healthy but untrained subjects as controls. We hypothesized that the cumulative effect of exercise training and the subsequent differences in muscle phenotype seen in these subject groups would be detectable in gene expression profiles of resting muscle and that divergent training histories would highlight novel genes and unique gene clusters that contribute to the altered phenotype of physically active individuals.
MATERIALS AND METHODS
Twenty healthy males volunteered for this investigation. Seven were endurance-trained (ET) cyclists who had been participating in endurance training for 8 yr (cycling 250-600 km·wk−1). These subjects had no history of strength/resistance training. Six subjects were strength-trained (ST) power lifters who had been participating exclusively in strength/resistance training for 9 yr (three to four sessions per week). Training histories were obtained from self-recorded training diaries. The ET and ST groups were collectively called trained (TR) group. The final seven subjects were healthy controls (Con) that did not participate in any formal exercise (Table 1) and whose tissue had been banked at Monash and RMIT Universities for up to 12 months. All subjects provided informed written consent, and the study was approved by the Human Research Ethics Committee of RMIT University and Monash University Standing Committee on Ethics Research on Humans.
Peak oxygen uptake (V˙O2peak) was determined during an incremental maximal cycling test to volitional fatigue on and isokinetic Lode bicycle ergometer (Groningen, Netherlands) as previously described (9,32). Briefly, the protocol commenced at an intensity equivalent to 3.33 W·kg−1 of body mass (BM) for ET or 2.5 W·kg−1 for ST and Con. This workload was maintained for 150 s then increased by 50 W for the second workload and 25 W every 150 s thereafter until the subjects reached volitional fatigue, defined as the inability to maintain a cadence >60 rpm. Throughout the maximal test, breath-by-breath expired gas was passed through a flowmeter, an O2 analyzer, and a CO2 analyzer (Medgraphics, St Paul, MN), which were calibrated with a 3-L Hans-Rudolph syringe and gases of known concentration (4.00% CO2 and 16.00% O2; BOC gases, Australia). The flowmeter and the gas analyzers were interfaced to a computer that calculated minute ventilation, oxygen uptake (V˙O2), CO2 production (V˙CO2), and respiratory exchange ratios from conventional equations.
Maximal concentric and eccentric strengths were determined using seated leg extensions performed on a Kin-Com isokinetic dynamometer (Chattanooga, TN) as previously described (9). The subject's right leg was strapped to the actuator arm immediately superior to the lateral malleolus of the lower leg, with the lateral condyle of the femur visually aligned to the fulcrum of the actuator arm. Contractions were performed at 30°·s−1, and subjects were instructed to extend their leg and to resist the actuator arm with maximal effort during repetitions. Exercise range of motion was 85° with leg extension end point set at −5° from full extension. Real-time visual feed back was provided for all repetitions. Quadriceps strength was determined during a series of 3-repetition maximal (RM) leg extension sets with each set separated by 120 s. Each individual 1-RM was defined as peak (N·m) recorded during the concentric and the eccentric contraction phases of the test protocol. Unfortunately, we were unable to conduct this test on the control subjects, as these subjects had participated in several alternate studies that did not include isokinetic dynamometry.
Subjects reported to the laboratory after 24 h of rest and an overnight (8 h) fast. The final meal on the day before attending the laboratory was provided to each subject to standardize macronutrient intake, and it consisted of 3 g CHO·kg−1 BM, 0.5 g protein·kg−1 BM, and 0.3 g fat·kg−1 BM, which subjects were asked to finish before 2000 h on the night before attending the laboratory. After voiding and resting for 15 min in a supine position, local anesthesia (Xylocaine) was administered to skin and fascia. A small incision was made in the anaesthetized skin and muscle fascia after which a percutaneous muscle biopsy was taken from the vastus lateralis using a 5-mm Bergström muscle biopsy cannulae (Stille, Sweden) with suction applied (11). These biopsy samples were immediately frozen in liquid nitrogen. Samples were then stored at −80°C for later analysis.
Total RNA was extracted from 20 to 30 mg of muscle according to modified methods of Chomczynski and Sacchi (7). Briefly, samples were homogenized in 0.5 mL of Trizol reagent (Invitrogen Life Technologies, Australia) and incubated at room temperature for 5 min. After addition of chloroform (100 μL), samples were incubated for 3 min at room temperature and centrifuged for 15 min (12,000g, 4°C), and the aqueous phase, containing the RNA, was precipitated with an equal volume of 100% ethanol. Total RNA was further purified through a Qiagen RNeasy Microcolumn (Qiagen, Germany) according to manufacturer's protocol. RNA was resuspended in 10-μL RNAse-free water. RNA concentration and estimation of purity were determined by measuring the absorbance of each RNA sample at 260 and 280 nm on a Nanodrop Spectrophotometer (Eppendorph, Hamberg, Germany) yielding 0.31 ± 0.05, 0.29 ± 0.03, and 0.44 ± 0.06 μg·μL−1 for ET, ST, and Con, respectively. The purity was considered high as the mean 260/280-nm ratio was 1.95 ± 0.02. Universal human RNA was used as the normalizing or reference RNA for the gene array experiments and was purchased from Stratagene (Lot#0240515). All RNA samples were stored at −80°C until use.
Due to the small quantities of muscle RNA, both the muscle and the reference RNA were amplified using the MessageAmp™ aRNA Kit (Ambion, Austin, TX) according to the manufacturer's instructions. All RNA extractions from muscle samples were conducted in the same laboratory under the same conditions, and these samples, along with control RNA, were amplified in the same run using reagents from kits with the same batch and lot numbers to reduce any variability from the linear amplification process. The process of amplification for microarray analysis has been used successfully in previous studies (22). Briefly, 0.5 μg of total RNA was reverse transcribed to synthesize first strand cDNA, followed immediately by the formation of the second strand of cDNA, which is then purified. A 12-h in vitro transcription reaction was used to synthesize amplified RNA (aRNA), which was then purified and resuspended in 100 μL of RNAse-free water. aRNA was then ethanol precipitated by storing overnight at −20°C in two volumes of 100% ethanol and 0.1 volume of 3 M of sodium acetate. aRNA was subsequently centrifuged at 12,000g for 40 min at 4°C, the pellet washed twice with 70% ethanol, and then the aRNA resuspended in RNAse-free water to a final concentration of 5 mg·mL−1.
Twenty glass microarray slides spotted with 8000 cDNA sequences were purchased from the AGRF (Human 8K, WEHI, Melbourne, Australia). Each subject was allocated a microarray where 50-70 μg of their amplified muscle RNA with an equal quantity of amplified reference RNA was used for cohybridization. Fluorescently labeled cDNA were created by an indirect-labeling procedure. Sample and reference RNA were prepared for labeling using the Superscript™ III indirect labeling system (Invitrogen Life Technologies), where the aRNA was mixed with 1 μL spike RNA and 2 μL anchored oligo (dT)20 primer (2.5 mg·mL−1) in a total volume of 18 μL and incubated for 5 min at 70°C. Next, 6 μL of 5× first strand buffer, 1.5 μL of 0.1 M dithiothreitol, 1.5 μL of dNTP mix with amino-modified nucleotides, 1 μL RNAseOUT™ (40 U·μL−1), and 2 μL of Superscript™ III reverse transcriptase (400 U·μL−1) were incubated for 2 h at 46°C. The aRNA template was then removed by hydrolysis with 1 N NaOH at 70°C for 10 min and was neutralized with 0.3 N acetic acid.
The cDNA was purified using Qiagen polymerase chain reaction (PCR) purification columns (Qiagen), according to the manufacturer's instructions. Reference and sample cDNA were labeled with Cy3 or Cy5 (CyDye postlabeling reactive dye pack; Amersham Biosciences #RPN 5661), respectively, on the column for 1 h in the dark. Samples were washed with PE buffer, eluted with 80 μL H2O. This eluate is mixed with 400 μL of phosphate buffer and then both added to a new column, Cy3 (reference) then Cy5 (sample). This column was washed with PE buffer, and the pooled DNA was eluted with 60 μL H2O. The solution was dried in a heated vacuum, then the pellet was resuspended and mixed in 16.2 μL of H2O, 30 μL of fresh 2× hybridization buffer (500 μL of formamide, 500 μL of 10× standard saline citrate [SSC], and 20 μL of 10% sodium dodecyl sulphate [SDS]), 5 μL of Cot1 DNA (Invitrogen), 3.8 μL of PolyA (10 mg·mL−1; Sigma), and 5 μL of Salmon Sperm DNA (10 mg·mL−1; Invitrogen). The samples were heated to 100°C for 2 min and cooled to room temperature before the 60 μL of solution were applied to the glass microarray slide and coverslipped via capillary action. The arrays were placed in Corning microarray hybridization chambers, which were immersed in a water bath (42°C) for 16 h. Hybridized slides were washed at room temperature with 1× SSC and 0.2% SDS for 5 min, 0.1× SSC and 0.2% SDS for 5 min, and twice in 0.1× SSC for 5 min. After washing, the slides were dried by centrifugation at 1500g for 5 min.
Scanning and quantification
The microarray slides were scanned using a Molecular Dynamics GenePix 4000B UV laser microarray scanner (Amersham Pharmacia Biotech, Piscataway, NJ) interfaced with a personal computer using GenePix v 5.2 (Amersham Pharmacia Biotech) to generate gene lists with corresponding fluorescence intensity data.
Data were imported into Gene Spring (Silicon Genetics) version 7.1, and unless otherwise indicated, statistical analysis was performed using default settings in this software. Per chip-intensity-dependent normalization (LOWESS) was carried out for each array, and a group of 5962 genes was identified that satisfied the criteria of a mean signal intensity of at least 150 in at least 50% of the hybridizations. Parametric ANOVA (Welsh ANOVA) was performed on log2-transformed ratios (M value), muscle sample fluorescence divided by the control sample fluorescence of same spot on the same array, of the normalized data for these 5962 genes with Benjamini-Hochberg false discovery rate correction at P ≤ 0.05 to identify differentially expressed genes across three populations of athletes. These data in raw and normalized format were submitted to the NBCI Gene Expression Ominbus database for public access (GSE9405) according to the MIAME standards (6).
Q-PCR gene validation
An aliquot of extracted mRNA was used for Q-PCR on six genes to validate the microarray results. Specifically, the mRNA was diluted to 50 ng·μL−1 then reverse transcribed to cDNA using a commercially available reverse transcription kit (Applied Biosystems, Victoria, Australia), to a final concentration of 10 ng·μL−1. Q-PCR was performed using the MasterCycler ep Realplex 2 (Eppendorf). Duplicate reaction volumes (20 μL) contained 2.5× RealMasterMix Probe PCR master mix (5 PRIME, Quantum Scientific, Australia) forward and reverse primers, and Taqman Probes (Applied Biosystems; Table 2) and cDNA template (diluted 1:40) were aliquoted using the ep Motion 5070 (Eppendorf). Samples were run in duplicate for one cycle (95°C, 2 min) followed by 45 cycles (95°C, 15 s and 60°C, 60 s), and fluorescence emissions were measured after each cycle. All calculations for relative gene expression determined by Q-PCR were conducted using qBase (34), which incorporated primer amplification efficiencies and normalization with the housekeeping gene and a control sample.
Gene ontology analysis was conducted on the list of differentially expressed genes, and gene lists of normalized data using Gene Tools (Web-based software and databases created by NTNU; http://www.genetools.no) and NCBI databases.
Normalized data from the gene lists were then subjected to a simplified cluster analysis using current literature and gene ontologies. The genes were divided into functional groups for genes, including nuclear-encoded mitochondrial genes, TCA cycle, electron transport chain, glycolysis, carbohydrate metabolism, fat metabolism, muscle contraction, development, and structure, and protein biosynthesis. Gene expression or M value (log2 [fold change ratio]) for each of the selected genes was averaged and expressed as a mean for each group (ET, ST, or Con). These data were then analyzed using standard statistical tests (see below). Gene expression values of specific genes from individual athletes or controls were exclude if the A value (log2 of geometric mean of normalized fluorescent intensities; log2 [(fluorescence of test RNA × fluorescence of reference RNA) / 2]) was <4.5, which indicated poor hybridization. To investigate whether training (TR compared with Con) per se or mode of training (ST compared with ET) effected gene expression within a cluster, further analyses were conducted. Genes from individual clusters were sorted for changes in expression, either increased (up) or decreased (down) between TR versus Con (M value of ET + ST/2 compared with M value Con) and ET versus ST (M value of ET compared with M value of ST).
The data obtained from the gene cluster analysis were further analyzed to investigate any possible relationships between the specific clusters or subgroups within clusters and the functional measures of muscle function (either V˙O2peak or peak concentric torque).
Subject characteristics, gene cluster analysis, and differentially expressed genes were compared using a one-way ANOVA. Contrasts between the groups were then determined using Tukey's post hoc tests. The relationships between any gene cluster or subgrouping and functional muscle parameters were analyzed by linear regression. These statistical analyses were conducted using GraphPad Prism ™ version 4 (GraphPad Software Inc., San Diego, CA). All data are expressed as mean ± SEM unless otherwise stated, and significance was accepted as P ≤ 0.05.
ET had a higher V˙O2peak compared with both the ST and the Con (Table 1; P < 0.05). Conversely, the ST athletes had higher eccentric and concentric peak forces during isokinetic knee extension compared with ET (Table 1; P < 0.05).
Microarrays and differential gene expression
Two hundred and sixty-three genes were shown to exhibit differential expression profiles in subjects that have been involved in exercise training for >8 yr (Tables 3 and 4). Of the 263 genes, 21 genes showed differential expression between ET and ST (Table 5). When these 263 genes were clustered into functional groups (Table 6), TR-up-regulated genes (n = 161) included muscle structure and development (n = 12), energy metabolism (n = 13), mitochondrial proteins (n = 13), protein handling (n = 13), and transcription and translation (n = 21). The 102 genes that were down-regulated in TR compared with Con muscle (Table 6) were in the same functional clusters but generally had fewer numbers of genes. The 21 genes differentially expressed between ST (n = 10) and ET (n = 11) were also clustered into functional groups (Table 6), with functional clusters and associated genes complimenting training history. Specifically, gene clusters for energy metabolism (n = 2) and mitochondrial proteins (n = 2) were up-regulated in ET compared with transcription and translation (n = 3) and protein handling (n = 2) in ST. Lacking information on the smallest physiologically relevant differences in basal gene expression between phenotypes, we conducted a simple cluster analysis using gene ontologies to cluster genes into functional groups.
To confirm the accuracy of the microarray data, we examined the relative gene expression of six genes differentially expressed in either ST or ET by Q-PCR (Table 7). Expression patterns were similar between the microarray and the Q-PCR measures for all genes measured (Table 7; P < 0.05). We selected up-regulated genes important for the specificity of training adaptation in ST (protein translation, proliferation, and structure) and ET (fuel metabolism).
Muscle contraction, development, and strength
This cluster of 157 genes was further subdivided into clusters of genes up- or down-regulated by training status (TR vs Con) or training mode (ET vs ST). In genes that were up-regulated by training, ET and ST subjects were not different but showed significantly greater expression than Con (Fig. 1B, P < 0.01). In those genes that were down-regulated by training, there were only small differences between ET and ST, but a significant difference was found when TR was compared with Con (Fig. 1B, P < 0.05).
When training groups were combined, changes in gene expression for the "muscle contraction, development and structure" gene cluster was not correlated to V˙O2peak or peak concentric force. However, in ST this up-regulated muscle protein cluster demonstrated a positive linear relationship with peak concentric torque (Fig. 2A; n = 10; r = 0.7; P < 0.05), whereas the same subcluster up-regulated by ET had a positive linear relationships with V˙O2peak (Fig. 2B; n = 19; r = 0.6; P < 0.01).
Mitochondrial genes of nuclear origin
ET significantly increased the expression of these genes compared with ST and Con (Fig. 3A; n = 327; P < 0.01). Further analysis showed that subclusters of genes were significantly increased or decreased by training mode (ET vs ST; Fig. 3C) and status (TR vs Con; Fig. 3B). Of these subclusters, ET showed significantly increased expression compared with ST and Con (Fig. 3C; n = 238; P < 0.01). Conversely, the subcluster of mitochondrial genes that were up-regulated by ST compared with ET and Con was not significant (Fig. 3C, closed bars). When training modes were combined (TR; Fig. 3B, open bars), both ET and ST subjects showed higher gene expression than Con (Fig. 3B; n = 201; P < 0.01). Similarly, genes that were down-regulated by TR were lower compared with Con (Fig. 3B, closed bars), but only tissue from ST individuals showed a significant decline in this subdivision of the cluster (Fig. 3B; n = 132; P < 0.01). In addition, M values of the nuclear-encoded mitochondrial gene cluster were related to V˙O2peak (Fig. 4A; r = 0.7; P < 0.001).
An additional subset of mitochondrial genes were selected from the array, which included genes encoding proteins of the electron transport chain (ETC; Fig. 3D-F). Training status increased the expression of the ETC gene cluster (Fig. 3E, open bars), but only ET reached significance (P < 0.05). When individual genes from the ETC cluster that were regulated by training mode were examined, ET showed significantly greater expression (Fig. 3F, open panels; P ≤ 0.05) compared with ST and Con. The ETC gene cluster up-regulated by ST (Fig. 3F, closed panels) demonstrated no differences between any of the groups.
Fat and carbohydrate metabolism
Analysis of all genes in the "carbohydrate and fat" regulatory clusters showed no difference between groups (Fig. 5A and C). When the carbohydrate cluster was subdivided into groups of genes up- or down-regulated by TR (Fig. 5B, open bars and closed bars, respectively), both ST and ET showed marked increases (open bars) and decreases (closed bars) in expression when compared with Con (Fig. 5B; P < 0.05). When fat regulatory genes were compared in the same manner, there was a strong tendency for an increase in expression in ET (Fig. 5D, open bars; P = 0.06). When these metabolic gene clusters were analyzed for relationships with muscle aerobic capacity, both fat metabolism (Fig. 4B; P < 0.01) and carbohydrate metabolism (Fig. 4C; P < 0.05) gene clusters had significant linear relationships with V˙O2peak (r = 0.65 and 0.54, respectively), whereas there was no significant relationship found between peak concentric torque and fat oxidative genes (r = −0.47, P = 0.17).
We analyzed all genes through manual clustering and found no difference in expression between groups (data not shown). However, breakdown into genes that were up-regulated by TR showed significantly greater expression than Con (Fig. 6A, open bars). Conversely, genes down-regulated by TR were more highly expressed in Con. Regression analyses of the protein biosynthesis gene cluster with performance measures showed nonsignificant tendencies toward a relationship with these measures, where genes in the cluster up-regulated by ST and ET correlated with peak concentric power (r = 0.57, P = 0.08) and V˙O2peak, respectively (r = 0.45, P = 0.06).
Adaptation of skeletal muscle to chronic endurance or strength training induces distinctly different phenotypes (8). We hypothesized that the cumulative effect of repeated exercise bouts (i.e., training) would result in detectable, coordinated changes in the muscle transcriptome, and that the specific contractile overload induced by different modes of exercise would reflect the distinct adaptive profiles and divergent training-induced phenotypes. We provide novel information regarding the basal transcriptional activity of well-trained athletes compared with healthy but untrained control subjects and report a similar pattern of changes in basal gene expression with training independent of the mode of activity undertaken by athletes from divergent training backgrounds. In addition, we identify novel genes that are differentially expressed in ET versus ST athletes (Table 6) as well as in the trained compared with untrained individuals (Tables 3 and 4).
Chronic training and gene expression
The primary finding of the present study is that exercise training results in an alteration in the pattern of transcription within skeletal muscle, indicative of a new genetic "set point" compared with sedentary individuals. This is highlighted through the initial microarray analysis that identified ∼260 genes (Tables 3 and 4) whose expression levels were more than 1.5-fold different compared with untrained subjects, regardless of the training history. Specifically, 161 genes were up-regulated by exercise training (Table 6), all of which were from gene clusters that are functionally relevant for skeletal muscle homeostasis. Knowing that the proteins encoded by these genes have well-defined functions in skeletal muscle and in some instances well-documented responses to training, it may be reasonable to suggest that the chronic alteration of these genes' expression significantly contributes to the increased performance capacity and health status of skeletal muscle after regular physical activity. It should be noted that we cannot discount a "natural selection" component to the trained individual's chosen sporting endeavor, whereby genetic elements predisposed these athletes toward specific activities. However, inclusion and resulting data from the untrained control group lend support to our conclusions.
To detect subtle changes in gene expression, we used a clustering approach to analyze our array data, adopting a more simplified manual clustering procedure. This was used as the data generated by the two-color cDNA microarray approach did not permit data entry into the GSEA cluster analysis program successfully used previously (24). Using this approach, we discovered functional gene clusters and subsets of genes within these clusters that were significantly regulated by training (Figs. 1, 3, 5, and 6). Specific changes identified that regular muscle contraction, whether prolonged and submaximal or short and intensive, significantly alters the expression of clusters of genes that control muscle contractile processes, structure, and development (Fig. 1) and carbohydrate metabolism (Fig. 5). Furthermore, the clusters of up-regulated genes correlated significantly with performance measures for both strength and endurance capacity (Figs. 2 and 4).
Exercise mode and gene expression
The small number of genes (n = 21; Tables 5 and 6) that were differentially expressed between ET and ST athletes was surprising. As such, the cluster analyses of genes in muscle-specific functional groups were more informative (Figs. 1-6). Therefore, the possibility exists that the phenotypic differences promoting functional divergence in skeletal muscle for resistance to fatigue (ET) versus strength and power (ST) may be the result of the interaction of individual genotype and merely subtle rather than substantial chronic changes in the transcriptome. Nonetheless, as might be expected, the genes chronically altered in ET subjects are predominantly involved in energy metabolism and mitochondrial function, membrane transport, and angiogenesis, whereas those up-regulated by prolonged strength training are involved in processes that regulate protein synthesis and transcription and translation (Table 6). Genes and gene clusters that control mitochondrial structure, function, and oxidative metabolism are globally and chronically up-regulated by endurance training only. Surprisingly there was only a trend toward chronic endurance training influencing genes and/or clusters for fat metabolism (Fig. 5D). Nevertheless, the fact that these correlated with V˙O2peak, but not peak concentric torque, suggests that endurance training favors long-term up-regulation of fat oxidative genes. Although these latter findings are not surprising based on previous evidence and the well-accepted importance of mitochondria in oxidative metabolism (10,26,30), we show for the first time that a great number of genes that regulate these processes are chronically elevated by endurance training and that this adaptation is specific to this type of training. When examining the subset of mitochondrial genes that were down-regulated by training (Fig. 3B), the expression level was dramatically lower in the ST subjects when compared sedentary controls. This observation supports the notion that strength training does not stimulate cell-signaling events that promote mitochondrial biogenesis and that specific adaptations to training may not be able to coexist with other (divergent) adaptations (3).
We expected to see significant links between the gene cluster regulating protein biosynthesis and the peak torque due to the link between muscle hypertrophy, protein synthesis, and strength training. However, the protein biosynthesis cluster of genes included many genes with divergent expression depending on training specificity. This functional cluster includes genes that control both cellular and mitochondrial biosynthesis (data not shown), and as such, the different training modes influenced the respective group of genes (Fig. 6). This finding was corroborated by the mean gene expression of the divergent subclusters up-regulated by either ET or ST correlated with V˙O2peak or peak concentric torque, respectively. This suggests that there are aspects of protein biosynthesis that are important for specific adaptive processes and, of note, supports the rest of the data in that there are chronic alterations in gene expression as a result of specific training modes. Although this may be the case, we do not imply that protein biosynthesis is increased in resting muscle of trained athletes, merely that the adaptation to training appears to include a chronically up-regulated capacity to increase protein biosynthesis upon receiving a stimulus. With respect to the importance of growth signaling and hypertrophy in response to strength training, these processes are strongly regulated by posttranslational modifications to transcription factors and kinases that control protein biosynthesis (3,9,23); thus, the long-term, low-level regulation of multiple genes in ST individuals might not be a critical factor for the chronic adaptations seen in these athletes.
ST individuals are characterized by muscle hypertrophy/protein synthesis, which is reflected in the novel genes and the gene clusters up-regulated compared with ET. These data support the notion that regular strength training chronically alters skeletal muscle gene expression to promote protein synthesis and hypertrophy, thus maintaining the phenotype for improved muscle force production. In contrast, the ET phenotype has a strong association with aspects related to mitochondrial biogenesis, improved oxygen supply, and energy metabolism. The novel genes and the gene clusters highlight the chronic up-regulation of genes specific to the characteristic phenotype for efficient energy provision from metabolism during rhythmic, continuous muscle contractions in endurance exercise.
There are several limitations of the present study that should be considered when interpreting the data set. First, subject numbers are relatively low, although in line with other studies that use similar approaches (9,19,22,31,36). Importantly, previous investigations report similar group sizes are sufficient to detect meaningful differences in gene profiles while also avoiding type 2 errors. This is likely because the profiles of the athletes in these investigations are so divergent. The timing of the postexercise muscle biopsy in the present study should also be considered. A previous microarray study, including reverse transcriptase polymerase chain reaction (RT-PCR) validation, reported that a small number of important metabolic regulatory genes remain elevated 48 h after high intensity endurance exercise, although the overwhelming majority of genes measured in that study had returned to baseline after 48 h. No sampling was performed 24 h after the cessation of exercise in that study (22), so direct comparisons with the present investigation are impossible. However, it should be noted that Yang et al. (37) measured selected genes 24 h after a bout of either endurance or resistance exercise in a similar subject cohort and reported that all genes under investigation had returned to resting levels after this time. The subjects in the current study were well conditioned, with a prolonged training history in their chosen sports. As such, we have previously shown that prior training history results in lesser metabolic disturbance and attenuated signaling responses compared with when subjects undertake an acute bout of exercise in the untrained state (9). One might therefore argue that the skeletal muscle of such well-trained subjects responds to a bout of familiar exercise and rapidly returns to baseline conditions within a short period (i.e., 24 h).
Perhaps of major importance is that we cannot rule out whether the genetic profiles of the divergent phenotypes under investigation are a direct result of the specific training stimulus or, in fact, represent "natural selection." In other words, we cannot rule out that our subjects represent extreme phenotypes that gravitate toward their respective sporting events through natural intrinsic abilities rather than training per se. This would also imply that skeletal muscle fiber type differences are present and would influence gene expression patterns accordingly. Because muscle fiber type was not determined in the present study, we can only speculate on this issue. As such, future studies examining gene expression in different subject cohorts should, if possible, include muscle fiber type analyses.
In summary, we provide evidence to support the notion that basal gene expression patterns in human skeletal muscle are altered after chronic (>8 yr) endurance and strength training. In particular, training-specific differences exist in groups of genes that are involved in adaptive processes that are important in response to different types of training. On the basis of this perspective, it is probable that chronic training subtly regulates numerous genes from important functional groups in human skeletal muscle, possibly through modulation of processes that control mRNA degradation, as a long-term adaptive mechanism to cope with repeated training stimuli. Whether this altered genetic "set point" remains indefinitely or gradually falls away in response to detraining is unknown. The findings from this study support the concept that chronic training adaptations include changes to the muscle transcriptome while at rest and that this probably contributes to improvement in performance characteristics of the muscle and therefore whole-body homeostasis during training-specific exercise.
The authors would like to thank Prof. Peter Rogers, Prof. Peter Fuller, Dr. Georgina Caruana, Leonie Cann, Babbette Fahey, and Maria Alexiadis for laboratory time, equipment access, and technical assistance. This work was supported by Monash University Strategic Grant Scheme 2005, Young Investigator (#ECR015); N.K. Stepto and a research grant from the Australian Sports Commission (to JAH and VC). ALC is supported by a Peter Doherty postdoctoral research fellowship from the National Health and Medical Research Council of Australia. The results of the present study do not constitute endorsement by ACSM.
1. Adhihetty PJ, Irrcher I, Joseph AM, Ljubicic V, Hood DA. Plasticity of skeletal muscle mitochondria in response to contractile activity. Exp Physiol
2. Arkinstall MJ, Tunstall RJ, Cameron-Smith D, Hawley JA. Regulation of metabolic genes in human skeletal muscle by short-term exercise and diet manipulation. Am J Physiol Endocrinol Metab
3. Baar K. Training for endurance and strength: lessons from cell signaling. Med Sci Sports Exerc
4. Booth FW, Chakravarthy MV, Spangenburg EE. Exercise and gene expression: physiological regulation of the human genome through physical activity. J Physiol (Lond)
5. Booth FW, Thomason DB. Molecular and cellular adaptions of muscle in response to exercise: perspectives of varoius models. Physiol Rev
6. Brazma A, Hingamp P, Quackenbush J, et al. Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nat Genet
7. Chomczynski P, Sacchi N. The single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction: twenty-something years on. Nat Protoc
8. Coffey VG, Hawley JA. The molecular bases of training adaptation. Sports Med
9. Coffey VG, Zhong Z, Shield A, et al. Early signaling responses to divergent exercise stimuli in skeletal muscle from well-trained humans. FASEB J
10. Costill DL, Fink WJ, Pollock ML. Muscle fiber composition and enzyme activities of elite distance runners. Med Sci Sports Exerc
11. Evans WJ, Phinney SD, Young VR. Suction applied to a muscle biopsy maximizes sample size. Med Sci Sports Exerc
12. Fluck M. Functional, structural and molecular plasticity of mammalian skeletal muscle in response to exercise stimuli. J Exp Biol
13. Fluck M, Dapp C, Schmutz S, Wit E, Hoppeler H. Transcriptional profiling of tissue plasticity: role of shifts in gene expression and technical limitations. J Appl Physiol
14. Fluck M, Hoppeler H. Molecular basis of skeletal muscle plasticity-from gene to form and function. Rev Physiol Biochem Pharmacol
15. Hawley JA. Adaptations of skeletal muscle to prolonged, intense endurance training. Clin Exp Pharmacol Physiol
16. Hawley JA, Stepto NK. Adaptations to training in endurance cyclists: implications for performance. Sports Med
17. Hawley JA, Zierath JR. Integration of metabolic and mitogenic signal transduction in skeletal muscle. Exerc Sport Sci Rev
18. Hickson RC. Interference of strength development by simultaneously training for strength and endurance. Eur J Appl Physiol Occup Physiol
19. Hittel DS, Kraus WE, Tanner CJ, Houmard JA, Hoffman EP. Exercise training increases electron and substrate shuttling proteins in muscle of overweight men and women with the metabolic syndrome. J Appl Physiol
20. Holloszy JO, Rennie MJ, Hickson RC, Conlee RK, Hagberg JM. Physiological consequences of the biochemical adaptations to endurance exercise. Ann N Y Acad Sci
21. Hoppeler H, Fluck M. Plasticity of skeletal muscle mitochondria: structure and function. Med Sci Sports Exerc
22. Mahoney DJ, Parise G, Melov S, Safdar A, Tarnopolsky MA. Analysis of global mRNA expression in human skeletal muscle during recovery from endurance exercise. FASEB J
23. Nader GA, Esser KA. Intracellular signaling specificity in skeletal muscle in response to different modes of exercise. J Appl Physiol
24. Patti ME, Butte AJ, Crunkhorn S, et al. Coordinated reduction of genes of oxidative metabolism in humans with insulin resistance and diabetes: potential role of PGC1 and NRF1. PNAS
25. Pilegaard H, Ordway GA, Saltin B, Neufer PD. Transcriptional regulation of gene expression in human skeletal muscle during recovery from exercise. Am J Physiol Endocrinol Metab
26. Puntschart A, Claassen H, Jostarndt K, Hoppeler H, Billeter R. mRNAs of enzymes involved in energy metabolism and mtDNA are increased in endurance-trained athletes. Am J Physiol
27. Rennie MJ, Tipton KD. Protein and amino acid metabolism during and after exercise and the effects of nutrition. Ann Rev Nutr
28. Roth SM, Ferrell RE, Peters DG, Metter EJ, Hurley BF, Rogers MA. Influence of age, sex, and strength training on human muscle gene expression determined by microarray. Physiol Genomics
29. Saltin B. Oxygen uptake and cardiac output during maximal treadmill and bicycle exercise. Mal Cardiovasc
30. Schmitt B, Fluck M, Decombaz J, et al. Transcriptional adaptations of lipid metabolism in tibialis anterior muscle of endurance-trained athletes. Physiol Genomics
31. Schmutz S, Dapp C, Wittwer M, Vogt M, Hoppeler H, Fluck M. Endurance training modulates the muscular transcriptome response to acute exercise. Pflugers Arch
32. Stepto NK, Martin DT, Fallon KE, Hawley JA. Metabolic demands of intense aerobic interval training in competitive cyclists. Med Sci Sports Exerc
33. Timmons JA, Larsson O, Jansson E, et al. Human muscle gene expression responses to endurance training provide a novel perspective on Duchenne muscular dystrophy. FASEB J
34. Vandesompele J, De Preter K, Pattyn F, et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol
35. Williamson DL, Gallagher PM, Carroll CC, Raue U, Trappe SW. Reduction in hybrid single muscle fiber proportions with resistance training in humans. J Appl Physiol
36. Wittwer M, Billeter R, Hoppeler H, Fluck M. Regulatory gene expression in skeletal muscle of highly endurance-trained humans. Acta Physiol Scand
37. Yang Y, Creer A, Jemiolo B, Trappe S. Time course of myogenic and metabolic gene expression in response to acute exercise in human. J Appl Physiol
38. Zierath JR, Hawley JA. Skeletal muscle fiber type: influence on contractile and metabolic properties. PLoS Biol
Keywords:©2009The American College of Sports Medicine
MICROARRAY; CLUSTER ANALYSIS; Q-PCR; RT-PCR; RESISTANCE TRAINING; AEROBIC TRAINING